Optical heterodyne detection systems involve mixing an input signal with a local oscillator signal and detecting the resulting beat frequency. Optical heterodyne detection systems can be utilized for optical spectrum analysis of an input optical signal by mixing the input signal with a local oscillator signal that is swept across a range of wavelengths or frequencies. Heterodyne-based optical signal analyzers can provide very high resolution, and are used for monitoring and analyzing communication systems based on dense wavelength division multiplexing.
The heterodyne signal is highly dependent on the polarization of the light in both the unknown input signal and the local oscillator. Hence, a mechanism must be provided to remove this source of variability from optical heterodyne-based optical receivers. If the polarization of the local oscillator is constant over time and does not vary with frequency while the oscillator is swept, a polarization diverse receiver can be achieved by dividing both the input signal and the local oscillator into two orthogonal polarization components and measuring the amplitude of the beat frequency obtained with each of the orthogonal polarizations. The results of the measurements at each polarization are then combined.
Such polarization diverse receivers utilize a polarization rotator and a walk-off crystal to generate four optical signals that provide the optical signals that are combined to produce the polarization diverse measurement. Two of these signals correspond to one of the orthogonal polarization directions, and the other two signals provide the data for the other orthogonal polarization direction. These receivers assume that the polarization of the local oscillator signal is fixed. The polarization rotator rotates the polarization of the local oscillator such that the local oscillator signal is divided equally between each of the orthogonal polarization directions. If the polarization direction of the local oscillator changes during the measurements, the amplitude of the local oscillator signal to measure the unknown signal at one polarization will be different than that used to measure the unknown signal at the orthogonal polarization. The algorithm used to combine the two polarization measurements depends on the amplitude of the local oscillator signal at one polarization having a known relationship to the amplitude of the local oscillator signal used to measure the unknown signal component at the orthogonal polarization. Hence, such variations result in errors in the measurement of the unknown signal.
A second problem with such polarization diverse receivers relates to the cost of the polarization rotator. If a waveplate is utilized to rotate the polarization, the waveplate must operate over a significant range of wavelengths, which substantially increases the cost of the waveplate. Alternatively, a Faraday rotator can be utilized to rotate the polarization; however, this requires a substantial amount of space and a magnetic field generator, which also increases the cost of the analyzer.
FIG. 1 illustrates a prior art optical heterodyne receiver. The input signal on optical fiber 21 is mixed with the local oscillator signal on optical fiber 22 by a polarization maintaining waveguide coupler 11 to generate two mixed signals. It is assumed that the input optical fiber that supplies the local oscillator signal is a polarization maintaining fiber.
The polarization of the signal on optical fiber 21 is not known, and hence, must be assumed to be different from that of the local oscillator signal (LO). Accordingly, the two mixed signals are separated into two pairs of signals in which each pair has the same polarization. The local oscillator is assumed to have a polarization as shown by arrow 31. The polarization of each of the mixed signals is rotated by 45 degrees by polarization rotator 12. The two mixed signals are then input to a walk-off crystal 13 that separates each signal into two signals having orthogonal polarizations that are separated in space. It is assumed that the axis of walk-off crystal 13 is aligned with that of the polarization of the LO signal. The resulting signals are shown at 23-26. A grin lens 14 images these signals into four corresponding optical fibers 27-30 that apply each signal to a corresponding photodiode in receiver 16. A signal processor 266 processes the signals from these photodiodes. The optical fibers 27-30 are maintained in the correct spatial positions relative to walk-off crystal 13 by a fiber holder 15.
Because the methods for processing the optical signals are known to the art, these methods will not be discussed in detail here. For the purposes of the present discussion, it is sufficient to note that the two signals corresponding to each polarization are subtracted from one another to remove the contributions that depend only on the intensity of the LO signal and the intensity of the unknown signal. The resultant two signals are then combined to provide a beat signal that is independent of the polarizations of the LO and input signal.
Receiver 10 has two problems. First, receiver 10 assumes that the polarization of the LO signal does not vary. If the polarization of the LO signal varies, an error will be introduced into the measured heterodyne signal because the LO signal will not be divided equally between the two orthogonal polarizations. Second, receiver 10 requires a polarization rotator to rotate the polarization of the mixed signals to the desired relationship with respect to the axis of walk-off crystal 13. This polarization rotator must operate over a wide range of optical frequencies, and hence, adds a significant cost to receiver 10.